Statistical Analsyses Code Lock et al NCB 2018
Lock et al
2018-08-17
- Setup
- Figure 1
- Figure 2
- Figure 3
- Figure 5
- Import Data for Figure 5
- Analyse Primary siRNA Screen responses
- Analyse Neomycin and LY drug treatment responses
- Boxplots of b5 adhesion intensity changes due to neomycin or LY
- Mann-Whitney test for significant Drug treatment effects on Reticular adhesion intensity
- Mann-Whitney test for significant Drug treatment effects on Focal adhesion intensity
- Mann-Whitney test for significant Drug treatment effects on intensity ratio Z-score
- Figure 6
- Figure 7
- Figure 8
- Supp Fig 5
- Import Data Supp Fig 5
- Analyse Arp2/3 inhibitor responses relative to control
- Boxplot changes in reticular adhesion b5 intensity2
- Mann-Whitney test for significant Drug treatment effects on Reticular adhesion intensity2
- Mann-Whitney test for significant Drug treatment effects on Focal adhesion intensity2
- Mann-Whitney test for significant Drug treatment effects on intensity ratio Z-score2
- Supp Fig 6
**
This R markdown document (.Rmd) contains analysis, visualisation and statistical testing relating to research paper “Reticular adhesions mediate cell-matrix attachment during mitosis” by Lock et al (NCB, 2018).
This code uses the R markdown framework and enables production of statistical and graphical outputs equivalent to those presented in the paper. Some purely aesthetic differences exist to final paper figures given subsequent editing (adobe illustrator) for figure embedding. All R packages necessary to run this code will be installed automatically upon running this code.
This code calls the source data file named “Supplementary Table 1 Statistics Source Data.xlsx” and selects individual data sheets pertinent to each analysis. These sheets are named according the Figure Panel that they underpin. The user must place this source data file in a location of their choice and modify the “data_input_path” variable to reflect this location. Graphical outputs are saved according to the relevant Figure Panel names (with some additional descriptive information) to a location defined by the variable “panel_output_path”. By default this is in a folder named “Graphical_Outputs” within the defined “data_input_path”. Note that some figure panels were generated directly within Excel and these can be found within the source data file.
In addition to specific .svg graphical outputs, this R markdown code will produce an HTML notebook that is useful to collectively preview and efficiently share outputs and their underlying code. This output HTML notebook file matches the name of this source .Rmd file, i.e. “Statistical_Analysis_Code_Lock_et_al_NCB_2018.nb.html”.
**
Setup
Input Path
rr # Set your path to input data data_input_path <- /Users/johnlock/Dropbox/A-CMAC Manchester collaboration/NCB submission documents/Resubmission 2/Source_Data
# name of the source data file which you have placed in the above location data_file <- Table 1 Statistics Source Data.xlsx
Output Path
rr # Define panel output path panel_output_path <- paste(data_input_path, _Outputs/, sep = /)
Figure 1
Import Data for Figure 1
rr # Import individual data files Mass.Spec.Data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 1A, col_names = TRUE) b5.int.vs.VN.conc <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 1G, col_names = TRUE)
Mass Spec of Integrin subunits in long-term culture
Figure 1A boxplot of Integrin subunits spectral counts
rr #define panel_file_name Integrin_Spectral_Counts_File_Name <- 1A Integrin Spectral Counts Boxplot
boxplot(Mean.Spectral.Counts ~ Integrin.Subunit, data = Mass.Spec.Data, notch = FALSE, varwidth = FALSE, las=1, ylab = Spectral Counts, col = , ylim = c(0, 50), outline = FALSE, boxlty = 0, whisklty = 0, staplelty = 0, medlwd = 0.5, boxfill = NA) beeswarm(Mean.Spectral.Counts ~ Integrin.Subunit, data = Mass.Spec.Data, method = ‘center’, add = T, col = , pch = 19, cex = 0.7) r svglite(paste(panel_output_path, Integrin_Spectral_Counts_File_Name, .svg, sep = \), width=8, height=6) boxplot(Mean.Spectral.Counts ~ Integrin.Subunit, data = Mass.Spec.Data, notch = FALSE, varwidth = FALSE, las=1, ylab = Spectral Counts, col = , ylim = c(0, 50), outline = FALSE, boxlty = 0, whisklty = 0, staplelty = 0, medlwd = 0.5, boxfill = NA) r beeswarm(Mean.Spectral.Counts ~ Integrin.Subunit, data = Mass.Spec.Data, method = ‘center’, add = T, col = , pch = 19, cex = 0.7) dev.off()
quartz_off_screen
2 
T-testing for spectral count differences differences
rr AVvsB1 <- subset(Mass.Spec.Data, Integrin.Subunit == | Integrin.Subunit == 1) T_test_AVvsB1 <- t.test(Mean.Spectral.Counts ~ Integrin.Subunit, data = AVvsB1, alternative = .sided, paired = FALSE)$p.value T_test_AVvsB1
[1] 0.005444416rr B5vsB1 <- subset(Mass.Spec.Data, Integrin.Subunit == 5 | Integrin.Subunit == 1) T_test_B5vsB1 <- t.test(Mean.Spectral.Counts ~ Integrin.Subunit, data = B5vsB1, alternative = .sided, paired = FALSE)$p.value T_test_B5vsB1
[1] 0.0001080229Integrin intensity responses to Vitronection concentration in talin-positive and talin-negative adhesions
Boxplot of Integrin intensity vs Vitronection concentration in talin-positive and talin-negative adhesions
rr #define panel_file_name b5_Intensity_File_Name <- 1G Integrin b5 Intensity
names(b5.int.vs.VN.conc) <- c(_Pos_vs_Neg, , .integrin.b5.Intensity) b5.int.vs.VN.conc\(Talin_Pos_vs_Neg <- recode(b5.int.vs.VN.conc\)Talin_Pos_vs_Neg, Atypical = _Neg) b5.int.vs.VN.conc\(Talin_Pos_vs_Neg <- recode(b5.int.vs.VN.conc\)Talin_Pos_vs_Neg, typical = _Pos) b5.int.vs.VN.conc\(Group <- paste(b5.int.vs.VN.conc\)Talin_Pos_vs_Neg, b5.int.vs.VN.conc\(Condition, sep = \_\) b5.int.vs.VN.conc\)Group <- factor(b5.int.vs.VN.conc$Group, levels = c(_Pos_1VN, _Neg_1VN, _Pos_3VN, _Neg_3VN, _Pos_10VN, _Neg_10VN)) ggplot(data = b5.int.vs.VN.conc, aes(x = Group, y = Mean.integrin.b5.Intensity)) + geom_boxplot(aes(fill = Talin_Pos_vs_Neg, ymin=..lower.., ymax=..upper..), outlier.shape=NA, notch = TRUE) + scale_fill_manual(values = c(_Pos = #4F80BC, _Neg = #C0504F)) + theme_classic() + coord_cartesian(ylim = c(500, 1300)) + theme(axis.text.x = element_text(angle = 45, hjust = 1)) svglite(paste(panel_output_path, b5_Intensity_File_Name, .svg, sep = \), width=5, height=8) r ggplot(data = b5.int.vs.VN.conc, aes(x = Group, y = Mean.integrin.b5.Intensity)) + geom_boxplot(aes(fill = Talin_Pos_vs_Neg, ymin=..lower.., ymax=..upper..), outlier.shape=NA, notch = TRUE) + scale_fill_manual(values = c(_Pos = #4F80BC, _Neg = #C0504F)) + theme_classic() + coord_cartesian(ylim = c(500, 1300)) + theme(axis.text.x = element_text(angle = 45, hjust = 1)) dev.off()
quartz_off_screen
2 
Wilcoxon Rank Sum testing of b5 intensity changes given increased Vitronectin
rr TalinPos = subset(b5.int.vs.VN.conc, b5.int.vs.VN.conc\(Talin_Pos_vs_Neg == \Talin_Pos\) TalinPos_VN1_vs_VN3 <- subset(b5.int.vs.VN.conc, b5.int.vs.VN.conc\)Talin_Pos_vs_Neg == _Pos & b5.int.vs.VN.conc\(Condition == \1VN\ | b5.int.vs.VN.conc\)Talin_Pos_vs_Neg == _Pos & b5.int.vs.VN.conc\(Condition == \3VN\ ) TalinPos_VN3_vs_VN10 <- subset(b5.int.vs.VN.conc, b5.int.vs.VN.conc\)Talin_Pos_vs_Neg == _Pos & b5.int.vs.VN.conc\(Condition == \3VN\ | b5.int.vs.VN.conc\)Talin_Pos_vs_Neg == _Pos & b5.int.vs.VN.conc\(Condition == \10VN\ ) Mu_Test_TalinPos_VN1_vs_VN3 = wilcox.test(Mean.integrin.b5.Intensity ~ Condition, data = TalinPos_VN1_vs_VN3)\)p.value Mu_Test_TalinPos_VN3_vs_VN10 = wilcox.test(Mean.integrin.b5.Intensity ~ Condition, data = TalinPos_VN3_vs_VN10)\(p.value TalinNeg_VN1_vs_VN3 <- subset(b5.int.vs.VN.conc, b5.int.vs.VN.conc\)Talin_Pos_vs_Neg == _Neg & b5.int.vs.VN.conc\(Condition == \1VN\ | b5.int.vs.VN.conc\)Talin_Pos_vs_Neg == _Neg & b5.int.vs.VN.conc\(Condition == \3VN\ ) TalinNeg_VN3_vs_VN10 <- subset(b5.int.vs.VN.conc, b5.int.vs.VN.conc\)Talin_Pos_vs_Neg == _Neg & b5.int.vs.VN.conc\(Condition == \3VN\ | b5.int.vs.VN.conc\)Talin_Pos_vs_Neg == _Neg & b5.int.vs.VN.conc\(Condition == \10VN\ ) Mu_Test_TalinNeg_VN1_vs_VN3 = wilcox.test(Mean.integrin.b5.Intensity ~ Condition, data = TalinNeg_VN1_vs_VN3)\)p.value Mu_Test_TalinNeg_VN3_vs_VN10 = wilcox.test(Mean.integrin.b5.Intensity ~ Condition, data = TalinNeg_VN3_vs_VN10)$p.value
Figure 2
Import Data for Figure 2
rr # Set your path to input data data_input_path <- /Users/johnlock/Dropbox/A-CMAC Manchester collaboration/NCB submission documents/Resubmission 2/Source_Data
data_file <- Table 1 Statistics Source Data.xlsx
# Import individual data files FRAP.data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 2S_T, col_names = TRUE) STORM.Data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 2V_W, col_names = TRUE)
Analysis of integrin b5 FRAP recovery curves in Reticular and Focal adhesions
Integrin b5 FRAP analysis comparing Reticular and Focal adhesions
rr #define panel_file_name FRAP_Curve_File_Name <- 2S b5 FRAP recovery
names(FRAP.data) <- c(_b5_FRAP_recovery_relative_to_prebleach, , _ROI, _vs_Focal) FRAP.data\(Reticular_vs_Focal <- recode(FRAP.data\)Reticular_vs_Focal, atypical = ) FRAP.data\(Reticular_vs_Focal <- recode(FRAP.data\)Reticular_vs_Focal, typical = ) ggplot(data=FRAP.data, aes(x=Time, y=Standardised_b5_FRAP_recovery_relative_to_prebleach, colour = Reticular_vs_Focal), alpha = 0.1) + stat_summary(fun.data =_cl_normal, alpha = 0.5) + theme_classic() + scale_colour_manual(values = c( = #BF4C49, = #4B7FBB)) + geom_smooth(method = , span = 0.2, size = 1, fill = #3A3A3A) svglite(paste(panel_output_path, FRAP_Curve_File_Name, .svg, sep = \), width=6, height=4) r ggplot(data=FRAP.data, aes(x=Time, y=Standardised_b5_FRAP_recovery_relative_to_prebleach, colour = Reticular_vs_Focal), alpha = 0.1) + stat_summary(fun.data =_cl_normal, alpha = 0.5) + theme_classic() + scale_colour_manual(values = c( = #BF4C49, = #4B7FBB)) + geom_smooth(method = , span = 0.2, size = 1, fill = #3A3A3A) dev.off()
quartz_off_screen
2 
KS test of FRAP recovery curves
rr Reticular.FRAP.data <- subset(FRAP.data, Reticular_vs_Focal == ) Focal.FRAP.data <- subset(FRAP.data, Reticular_vs_Focal == ) Reticular.FRAP.data.LOESS <- loess(Standardised_b5_FRAP_recovery_relative_to_prebleach ~ Time, data = Reticular.FRAP.data, span = 0.2) Focal.FRAP.data.LOESS <- loess(Standardised_b5_FRAP_recovery_relative_to_prebleach ~ Time, data = Focal.FRAP.data, span = 0.2) Reticular.FRAP.data.LOESS.singular <- unique(Reticular.FRAP.data.LOESS\(fitted) Focal.FRAP.data.LOESS.singular <- unique(Focal.FRAP.data.LOESS\)fitted) KS.Reticular.vs.Focal.FRAP <- ks.test(Reticular.FRAP.data.LOESS.singular, Focal.FRAP.data.LOESS.singular, alternative = .sided) print(KS.Reticular.vs.Focal.FRAP$p.value)
[1] 0.002107921Analysis of interphase STORM data
STORM data wrangling1
rr ## Determine CMAC.type (Reticular vs Focal depending on presence of A in \(File (column)) CMAC.type = c() for (cluster in 1:length(STORM.Data\)File) ) { CMAC.type[cluster] <- if(grepl(, STORM.Data\(File[cluster]) == TRUE) { \Reticular\
} else if(grepl(\T\, STORM.Data\)File[cluster]) == TRUE) {
} else if(grepl(, STORM.Data\(File[cluster]) == TRUE) { \Non-Retraction\
} else { \Retraction\
} } STORM.Data\)CMAC.type <- cbind(as.character(CMAC.type)) ## Unique nanocluster ID generation = merge of Folder, File and Cluster # STORM.Data\(Unique.ncID <- as.factor(paste(STORM.Data\)Folder, STORM.Data\(File, STORM.Data\)Cluster)) STORM.Data\(Unique.CMACID <- as.factor(paste(STORM.Data\)Folder, STORM.Data$File)) STORM.Data <- as.data.frame(STORM.Data)
Nearest Neighbour Analysis between localisations within unique nanoclusters
rr ## NND analysis of Incite3 data NNND <- c() for (i in 1:length(unique(STORM.Data\(Unique.CMACID)) ) { temp = STORM.Data[STORM.Data\)Unique.CMACID == unique(STORM.Data$Unique.CMACID)[i], ] NNND = append(NNND, nndist(temp[,7], temp[,8], k=1)) }
NAs introduced by coercionNAs introduced by coercionrr STORM.Data$NNND <- cbind(NNND)
Boxplots summarising Nearest Neighbout Distance per nanocluster
rr NNND.boxplot.filename <- 2V nearest neighbour boxplot
NNND.A = subset(STORM.Data\(NNND, CMAC.type == \Reticular\) NNND.T = subset(STORM.Data\)NNND, CMAC.type == ) svglite(paste(panel_output_path, NNND.boxplot.filename, .svg, sep = \), width=6, height=4) boxplot(NNND ~ CMAC.type, subset(STORM.Data, CMAC.type == | CMAC.type == ), notch = TRUE, outline = FALSE, main = Neighbour Distance of Nanocluster by CMAC Type, xlab = type, ylab = Neighbour Distance (nm), plot =
) dev.off()
null device
1 rr boxplot(NNND ~ CMAC.type, subset(STORM.Data, CMAC.type == | CMAC.type == ), notch = TRUE, outline = FALSE, main = Neighbour Distance of Nanocluster by CMAC Type, xlab = type, ylab = Neighbour Distance (nm), plot =
)
Boxplots summarising Molecular Localization per nanocluster
rr Molecule.number.boxplot.filename <- 2W molecular number boxplot
Molecules.A = as.numeric(subset(STORM.Data\(Molecules, CMAC.type == \Reticular\)) Molecules.T = as.numeric(subset(STORM.Data\)Molecules, CMAC.type == )) svglite(paste(panel_output_path, Molecule.number.boxplot.filename, .svg, sep = \), width=6, height=4) boxplot(as.numeric(Molecules) ~ CMAC.type, subset(STORM.Data, CMAC.type == | CMAC.type == ), notch = TRUE, outline = FALSE, main = of Integrin avb5 Molecules by CMAC Type, xlab = type, ylab = of Integrin avb5 Molecules, plot =
) dev.off()
null device
1 rr boxplot(as.numeric(Molecules) ~ CMAC.type, subset(STORM.Data, CMAC.type == | CMAC.type == ), notch = TRUE, outline = FALSE, main = of Integrin avb5 Molecules by CMAC Type, xlab = type, ylab = of Integrin avb5 Molecules, plot =
)
Figure 3
Import Data for Figure 3
rr # Set your path to input data data_input_path <- /Users/johnlock/Dropbox/A-CMAC Manchester collaboration/NCB submission documents/Resubmission 2/Source_Data
data_file <- Table 1 Statistics Source Data.xlsx
# Import individual data files CytoD2h_vs_control <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 3C, col_names = TRUE) CytoD2h_vs_control <- CytoD2h_vs_control[,1:3] Adhesion.Assay.Data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 3E, col_names = TRUE) Adhesion.Assay.Data <- Adhesion.Assay.Data[,1:4] Talin.KD.Data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 3H_I, col_names = TRUE)
Adhesion Assay
Fig 3E Plot Adhesion Levels per Condition
rr oldpar <- par() par(mar = c(10, 2.5, 1, 1)) #define panel_file_name Adhesion_Assay_panel_file_name <- 3E Adhesion Assay Boxplots
Adhesion.Assay.Data\(Condition <- factor(Adhesion.Assay.Data\)Condition, unique(Adhesion.Assay.Data\(Condition)) max_adhesion <- max(Adhesion.Assay.Data\)NumberObjects) Adhesion.Assay.Data\(Normalised_Adhesion <- Adhesion.Assay.Data\)NumberObjects/max_adhesion svglite(paste(panel_output_path, Adhesion_Assay_panel_file_name, .svg, sep = \), width=7.7, height=6.5) boxplot(Normalised_Adhesion ~ Condition, data = Adhesion.Assay.Data, notch = FALSE, varwidth = FALSE, outline = F, las=2, ylab = Cell Number, col = c(1wt -Cyto = #CB2327, 1wt +Cyto +RGD = #CB2327, 1b5 -Cyto = #4775A3, 1b5 -Cyto +RAD = #4775A3, 1b5 -Cyto +RGD = #9EC0E5, 1b5 +Cyto = #4775A3, 1b5 +Cyto +RAD = #4775A3, 1b5 +Cyto +RGD = #9EC0E5), ylim = c(0,1.2)) beeswarm(Normalised_Adhesion ~ Condition, data = Adhesion.Assay.Data, method = ‘center’, add = T, col = , pch = 19, cex = 0.7) dev.off()
quartz_off_screen
2 rr boxplot(Normalised_Adhesion ~ Condition, data = Adhesion.Assay.Data, notch = FALSE, varwidth = FALSE, outline = F, las=2, ylab = Cell Number, col = c(1wt -Cyto = #CB2327, 1wt +Cyto +RGD = #CB2327, 1b5 -Cyto = #4775A3, 1b5 -Cyto +RAD = #4775A3, 1b5 -Cyto +RGD = #9EC0E5, 1b5 +Cyto = #4775A3, 1b5 +Cyto +RAD = #4775A3, 1b5 +Cyto +RGD = #9EC0E5), ylim = c(0,1.2)) beeswarm(Normalised_Adhesion ~ Condition, data = Adhesion.Assay.Data, method = ‘center’, add = T, col = , pch = 19, cex = 0.7)
T-testing for Adhesion Level differences with Holm-Bonferroni p-value correction
rr p_values <- c() for (i in 1:length(unique(Adhesion.Assay.Data\(Condition))){ for (j in 1:length(unique(Adhesion.Assay.Data\)Condition))){ Condition_1 = subset(Adhesion.Assay.Data, Adhesion.Assay.Data\(Condition == unique(Adhesion.Assay.Data\)Condition)[i]) Condition_1\(Condition_Number = as.factor(i) Condition_2 = subset(Adhesion.Assay.Data, Adhesion.Assay.Data\)Condition == unique(Adhesion.Assay.Data\(Condition)[j]) Condition_2\)Condition_Number = as.factor(j+100) Condition_Pair = rbind(Condition_1, Condition_2) Condition_Pair_ttest = t.test(Normalised_Adhesion ~ Condition_Number, data = Condition_Pair, paired = FALSE)\(p.value p_values = rbind(p_values, Condition_Pair_ttest) } } p_values <- as.vector(p_values) p_value_matrix <- matrix(p_values,nrow = length(unique(Adhesion.Assay.Data\)Condition)),ncol = length(unique(Adhesion.Assay.Data\(Condition))) columns <- as.vector(unique(Adhesion.Assay.Data\)Condition)) rows <- as.vector(unique(Adhesion.Assay.Data\(Condition)) p_value_dataframe <- as.data.frame(p_value_matrix, row.names = rows) names(p_value_dataframe) <- columns # print(p_value_dataframe) # Introduce Holm (also called Holm-Bonferroni) p-value correction (used because it gives same stringency against false positives (type 1 errors) with lower probability of false negatives (type 2 errors))) corrected_p_values <- p.adjust(p_values, method = \holm\, n = length(p_values)) corrected_p_value_matrix1 <- matrix(corrected_p_values,nrow = length(unique(Adhesion.Assay.Data\)Condition)),ncol = length(unique(Adhesion.Assay.Data\(Condition))) columns <- as.vector(unique(Adhesion.Assay.Data\)Condition)) rows <- as.vector(unique(Adhesion.Assay.Data$Condition)) corrected_p_value_dataframe <- as.data.frame(corrected_p_value_matrix1, row.names = rows) names(corrected_p_value_dataframe) <- columns print(corrected_p_value_dataframe)
Plot heatmap of corrected p_values
rr heatmap.2(log(corrected_p_value_matrix1), dendrogram = ‘none’, Rowv = FALSE, Colv = FALSE, labRow = rows, labCol = columns, srtCol = 45, cexRow = 1, cexCol = 1, margins = c(7, 9), symm = TRUE, revC = FALSE, breaks = c(log(1e-300), log(1e-10), log(1e-6), log(0.001), log(0.01), log(0.05), log(1)), rowsep = 1:nrow(corrected_p_value_matrix1), colsep = 1:ncol(corrected_p_value_matrix1), sepcolor = ‘darkgrey’, sepwidth = c(0.02,0.02), trace = ‘none’, col = c(, , , , , ), denscol = NULL, keysize = 1.5, key.title = NA)
Wilcoxon Rank Sum testing for Adhesion Level differences
rr U_values <- c() for (i in 1:length(unique(Adhesion.Assay.Data\(Condition))){ for (j in 1:length(unique(Adhesion.Assay.Data\)Condition))){ Condition_1 = subset(Adhesion.Assay.Data, Adhesion.Assay.Data\(Condition == unique(Adhesion.Assay.Data\)Condition)[i]) Condition_1\(Condition_Number = as.factor(i) Condition_2 = subset(Adhesion.Assay.Data, Adhesion.Assay.Data\)Condition == unique(Adhesion.Assay.Data\(Condition)[j]) Condition_2\)Condition_Number = as.factor(j+100) Condition_Pair = rbind(Condition_1, Condition_2) Condition_Pair_Utest = wilcox.test(Normalised_Adhesion ~ Condition_Number, data = Condition_Pair)$p.value U_values = rbind(p_values, Condition_Pair_Utest) } }
cannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiescannot compute exact p-value with tiesrr U_values <- as.vector(U_values) U_value_matrix <- matrix(U_values, nrow = length(unique(Adhesion.Assay.Data\(Condition)),ncol = length(unique(Adhesion.Assay.Data\)Condition))) columns <- as.vector(unique(Adhesion.Assay.Data\(Condition)) rows <- as.vector(unique(Adhesion.Assay.Data\)Condition)) U_value_dataframe <- as.data.frame(U_value_matrix, row.names = rows) names(U_value_dataframe) <- columns print(U_value_dataframe)
Talin KD effects
Plot and test Talin KD effects on cell area and integrin b5 intensity within adhesions
rr # beeswarm(Total.Talin.per.Cell ~ siRNA..Oligo, data = Talin.KD.Data, method = ‘center’, col = , pch = 19, cex = 1, ylim = c(0,1)) # # beeswarm(Cell.Area ~ siRNA..Oligo, data = Talin.KD.Data, method = ‘center’, col = , pch = 19, cex = 1, ylim = c(0,1)) # # beeswarm(Mean.Clustered.b5.per.Cell ~ siRNA..Oligo, data = Talin.KD.Data, method = ‘center’, col = , pch = 19, cex = 1)#, ylim = c(0,1)) plot(log2(Talin.KD.Data\(Total.Talin.per.Cell), Talin.KD.Data\)Cell.Area, ylim = c(0,1), pch = c(0, 0, 0, 0, 2, 21), col = c(, , , , , )) abline(lm(Talin.KD.Data\(Cell.Area ~ log2(Talin.KD.Data\)Total.Talin.per.Cell)), col = ) r lm.cell.area_vs_talin.per.cell <- lm(Talin.KD.Data\(Cell.Area ~ log2(Talin.KD.Data\)Total.Talin.per.Cell)) TalinKD_Cell_Area_panel_file_name <- 3H TalinKD_vs_Cell_Area
svglite(paste(panel_output_path, TalinKD_Cell_Area_panel_file_name, .svg, sep = \), width=3, height=3) plot(log2(Talin.KD.Data\(Total.Talin.per.Cell), Talin.KD.Data\)Cell.Area, ylim = c(0,1), pch = c(0, 0, 0, 0, 2, 21), col = c(, , , , , ), cex.axis = 0.5, cex.label = 0.5) r abline(lm(Talin.KD.Data\(Cell.Area ~ log2(Talin.KD.Data\)Total.Talin.per.Cell)), col = ) dev.off()
quartz_off_screen
2 
rr plot(Talin.KD.Data\(Total.Talin.per.Cell, log2(Talin.KD.Data\)Mean.Clustered.b5.per.Cell), xlim = c(0,1), ylim = c(0,4), pch = c(0, 0, 0, 0, 2, 21), col = c(, , , , , )) abline(lm(log2(Talin.KD.Data\(Mean.Clustered.b5.per.Cell) ~ Talin.KD.Data\)Total.Talin.per.Cell), col = ) r lm.b5.intensity_vs_talin.per.cell <- lm(log2(Talin.KD.Data\(Mean.Clustered.b5.per.Cell) ~ Talin.KD.Data\)Total.Talin.per.Cell) TalinKD_b5_Intensity_panel_file_name <- 3I TalinKD_vs_b5_Intensity
svglite(paste(panel_output_path, TalinKD_b5_Intensity_panel_file_name, .svg, sep = \), width=3, height=3) plot(Talin.KD.Data\(Total.Talin.per.Cell, log2(Talin.KD.Data\)Mean.Clustered.b5.per.Cell), xlim = c(0,1), ylim = c(0,4), pch = c(0, 0, 0, 0, 2, 21), col = c(, , , , , ), cex.axis = 0.5, cex.label = 0.5) r abline(lm(log2(Talin.KD.Data\(Mean.Clustered.b5.per.Cell) ~ Talin.KD.Data\)Total.Talin.per.Cell), col = ) dev.off()
quartz_off_screen
2 
Test CytoD effects on adhesion intensities
Wilcoxon rank sum test of CytoD effect on vinculin and b5 intensity in adhesions
rr names(CytoD2h_vs_control) <- c(_Intensity, 5_Intensity, ) Vinculin_Intensity_mu_test <- wilcox.test(Vinculin_Intensity ~ Condition, data = CytoD2h_vs_control)$p.value print(Vinculin_Intensity_mu_test)
[1] 1.785123e-166rr b5_Intensity_mu_test <- wilcox.test(b5_Intensity ~ Condition, data = CytoD2h_vs_control)$p.value print(b5_Intensity_mu_test)
[1] 1.624755e-99Figure 5
Import Data for Figure 5
rr # Set your path to input data data_input_path <- /Users/johnlock/Dropbox/A-CMAC Manchester collaboration/NCB submission documents/Resubmission 2/Source_Data
data_file <- Table 1 Statistics Source Data.xlsx
# Import individual data files Primary.siRNA.Screen.Data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 5B_C, col_names = TRUE) Neo.LY.Drug.Treatment.Data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 5D, col_names = TRUE)
Analyse Primary siRNA Screen responses
Boxplot changes in reticular to focal adhesion b5 intensity ratio due to siRNA
rr siRNA_boxplot_summary_filename <- 5B siRNA boxplot summary
cell.numbers.primary.screen <- count(Primary.siRNA.Screen.Data, siRNA_Target) count(Primary.siRNA.Screen.Data, siRNA_Target) r mean.cell.numbers.primary.screen <- summarise(cell.numbers.primary.screen, avg = mean(n), SD = sd(n), sum(n)) summarise(cell.numbers.primary.screen, avg = mean(n), SD = sd(n), sum(n))
rr ggplot(Primary.siRNA.Screen.Data, aes(x=siRNA_Target, y=Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score, color = factor(siRNA_Treatment_Type))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(,,_3Ctrl_Pool, -TARGETplus Non-targeting Control, Non-targeting Control, , 5, , 4KA, 4K2B, , 5K1B, 5K1C, 3C2A)) + theme(axis.text.x = element_text(angle=90)) + ylim(-3, 6)#+ svglite(paste(panel_output_path, siRNA_boxplot_summary_filename, .svg, sep = \), width=7, height=5) r ggplot(Primary.siRNA.Screen.Data, aes(x=siRNA_Target, y=Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score, color = factor(siRNA_Treatment_Type))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(,,_3Ctrl_Pool, -TARGETplus Non-targeting Control, Non-targeting Control, , 5, , 4KA, 4K2B, , 5K1B, 5K1C, 3C2A)) + theme(axis.text.x = element_text(angle=90)) + ylim(-3, 6)#+ dev.off()
quartz_off_screen
2 
Parallel coordinates plot of adhesion responses to siRNA
rr siRNA_parcoord_summary_filename <- 5C siRNA parcoord summary
Primary_screen_siRNA_subset <- subset(Primary.siRNA.Screen.Data, siRNA_Target == | siRNA_Target == | siRNA_Target == _3Ctrl_Pool | siRNA_Target == -TARGETplus Non-targeting Control | siRNA_Target == Non-targeting Control | siRNA_Target == | siRNA_Target == 4KA | siRNA_Target == 4K2B | siRNA_Target == | siRNA_Target == 5K1B | siRNA_Target == 5K1C | siRNA_Target == 3C2A) Primary_screen_siRNA_subset_grouped <- aggregate(Primary_screen_siRNA_subset[, c(11,13,18)], list(Primary_screen_siRNA_subset\(siRNA_Target), median) Primary_screen_siRNA_subset_grouped\)Group.1 <- as.factor(Primary_screen_siRNA_subset_grouped$Group.1) names(Primary_screen_siRNA_subset_grouped) <- c(_Target, _Reticular_Adhesion_Intensity, _Focal_Adhesion_Intensity, _Reticular_Focal_Adhesion_Intensity) svglite(paste(panel_output_path, siRNA_parcoord_summary_filename, .svg, sep = \), width=7, height=5) ggparcoord(Primary_screen_siRNA_subset_grouped, columns = c(2,3,4), groupColumn = 1, scale = , order = c(2,3,4)) + theme(axis.text.x = element_text(angle=60, hjust = 1, size = 5)) dev.off()
null device
1 rr ggparcoord(Primary_screen_siRNA_subset_grouped, columns = c(2,3,4), groupColumn = 1, scale = , order = c(2,3,4)) + theme(axis.text.x = element_text(angle=60, hjust = 1, size = 5))
Mann-Whitney test for significant siRNA treatment effects
rr siRNA.conditions.for.U.test <- subset(Primary.siRNA.Screen.Data, siRNA_Target != & siRNA_Target != & siRNA_Target != _3Ctrl_Pool & siRNA_Target != Non-targeting Control) p_values <- c() for (i in 1:length(unique(siRNA.conditions.for.U.test\(siRNA_Target))){ for (j in 1:length(unique(siRNA.conditions.for.U.test\)siRNA_Target))){ siRNA_Target_1 = subset(siRNA.conditions.for.U.test, siRNA.conditions.for.U.test\(siRNA_Target == unique(siRNA.conditions.for.U.test\)siRNA_Target)[i]) siRNA_Target_1\(siRNA_Target_Number = as.factor(i) siRNA_Target_2 = subset(siRNA.conditions.for.U.test, siRNA.conditions.for.U.test\)siRNA_Target == unique(siRNA.conditions.for.U.test\(siRNA_Target)[j]) siRNA_Target_2\)siRNA_Target_Number = as.factor(j+100) siRNA_Target_Pair = rbind(siRNA_Target_1, siRNA_Target_2) siRNA_Target_Pair_ttest = wilcox.test(Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score ~ siRNA_Target_Number, data = siRNA_Target_Pair)\(p.value p_values = rbind(p_values, siRNA_Target_Pair_ttest) } } p_values <- as.vector(p_values) p_value_matrix <- matrix(p_values,nrow = length(unique(siRNA.conditions.for.U.test\)siRNA_Target)),ncol = length(unique(siRNA.conditions.for.U.test\(siRNA_Target))) columns <- as.vector(unique(siRNA.conditions.for.U.test\)siRNA_Target)) rows <- as.vector(unique(siRNA.conditions.for.U.test\(siRNA_Target)) p_value_dataframe <- as.data.frame(p_value_matrix, row.names = rows) names(p_value_dataframe) <- columns # print(p_value_dataframe) # Introduce Holm (also called Holm-Bonferroni) p-value correction (used because it gives same stringency against false positives (type 1 errors) with lower probability of false negatives (type 2 errors))) corrected_p_values <- p.adjust(p_values, method = \holm\, n = length(p_values)) corrected_p_value_matrix2 <- matrix(corrected_p_values,nrow = length(unique(siRNA.conditions.for.U.test\)siRNA_Target)),ncol = length(unique(siRNA.conditions.for.U.test\(siRNA_Target))) columns <- as.vector(unique(siRNA.conditions.for.U.test\)siRNA_Target)) rows <- as.vector(unique(siRNA.conditions.for.U.test$siRNA_Target)) corrected_p_value_dataframe <- as.data.frame(corrected_p_value_matrix2, row.names = rows) names(corrected_p_value_dataframe) <- columns print(corrected_p_value_dataframe)
Plot heatmap of corrected p_values mann-whitney
rr heatmap.2(log(corrected_p_value_matrix2), dendrogram = ‘none’, Rowv = FALSE, Colv = FALSE, labRow = rows, labCol = columns, srtCol = 45, cexRow = 1, cexCol = 1, margins = c(7, 9), symm = TRUE, revC = FALSE, breaks = c(log(1e-300), log(1e-50), log(1e-10), log(1e-6), log(0.001), log(0.01), log(0.05), log(1)), rowsep = 1:nrow(corrected_p_value_matrix2), colsep = 1:ncol(corrected_p_value_matrix2), sepcolor = ‘darkgrey’, sepwidth = c(0.02,0.02), trace = ‘none’, col = c(, , , , , , ), denscol = NULL, keysize = 1.5, key.title = NA)
Analyse Neomycin and LY drug treatment responses
Boxplots of b5 adhesion intensity changes due to neomycin or LY
rr PIP_drug_boxplot_summary_filename <- 5D PIP drug boxplot summary
cell.numbers.NEO_LY.screen <- count(Neo.LY.Drug.Treatment.Data, Drug) count(Neo.LY.Drug.Treatment.Data, Drug) r mean.cell.numbers.NEO_LY.screen <- summarise(cell.numbers.NEO_LY.screen, avg = mean(n), SD = sd(n), sum(n)) summarise(cell.numbers.NEO_LY.screen, avg = mean(n), SD = sd(n), sum(n))
rr svglite(paste(panel_output_path, PIP_drug_boxplot_summary_filename, .svg, sep = \), width=7, height=5) p1 <- ggplot(Neo.LY.Drug.Treatment.Data, aes(x=Drug, y=Median.Mean..z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(, _10mM,_25mM)) + theme(axis.text.x = element_text(angle=45, vjust = 0.5)) + coord_cartesian(ylim=c(-3, 6)) + ylab(Reticular Adhesion Intensity) + theme(legend.position=) ## plot T-CMAC intensity Z-score ggplot2 boxplot - per Drug p2 <- ggplot(Neo.LY.Drug.Treatment.Data, aes(x=Drug, y=Median.Mean._T.CMACs.z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(, _10mM,_25mM)) + theme(axis.text.x = element_text(angle=45, vjust = 0.5)) + coord_cartesian(ylim=c(-3, 6)) + ylab(Focal Adhesion Intensity) + theme(legend.position=) ## plot intensity ratio Z-score ggplot2 boxplot - per Drug p3 <- ggplot(Neo.LY.Drug.Treatment.Data, aes(x=Drug, y=Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(, _10mM,_25mM)) + theme(axis.text.x = element_text(angle=45, vjust = 0.5)) + coord_cartesian(ylim=c(-3, 6)) + ylab(Reticular:Focal Adhesion Intensity) + theme(legend.position=) grid.arrange(p1, p2, p3, ncol = 3) dev.off()
null device
1 rr p1 <- ggplot(Neo.LY.Drug.Treatment.Data, aes(x=Drug, y=Median.Mean..z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(, _10mM,_25mM)) + theme(axis.text.x = element_text(angle=45, vjust = 0.5)) + coord_cartesian(ylim=c(-3, 6)) + ylab(Reticular Adhesion Intensity) + theme(legend.position=) ## plot T-CMAC intensity Z-score ggplot2 boxplot - per Drug p2 <- ggplot(Neo.LY.Drug.Treatment.Data, aes(x=Drug, y=Median.Mean._T.CMACs.z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(, _10mM,_25mM)) + theme(axis.text.x = element_text(angle=45, vjust = 0.5)) + coord_cartesian(ylim=c(-3, 6)) + ylab(Focal Adhesion Intensity) + theme(legend.position=) ## plot intensity ratio Z-score ggplot2 boxplot - per Drug p3 <- ggplot(Neo.LY.Drug.Treatment.Data, aes(x=Drug, y=Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(, _10mM,_25mM)) + theme(axis.text.x = element_text(angle=45, vjust = 0.5)) + coord_cartesian(ylim=c(-3, 6)) + ylab(Reticular:Focal Adhesion Intensity) + theme(legend.position=) grid.arrange(p1, p2, p3, ncol = 3)
Mann-Whitney test for significant Drug treatment effects on Reticular adhesion intensity
rr p_values <- c() for (i in 1:length(unique(Neo.LY.Drug.Treatment.Data\(Drug))){ for (j in 1:length(unique(Neo.LY.Drug.Treatment.Data\)Drug))){ Drug_1 = subset(Neo.LY.Drug.Treatment.Data, Neo.LY.Drug.Treatment.Data\(Drug == unique(Neo.LY.Drug.Treatment.Data\)Drug)[i]) Drug_1\(Drug_Number = as.factor(i) Drug_2 = subset(Neo.LY.Drug.Treatment.Data, Neo.LY.Drug.Treatment.Data\)Drug == unique(Neo.LY.Drug.Treatment.Data\(Drug)[j]) Drug_2\)Drug_Number = as.factor(j+100) Drug_Pair = rbind(Drug_1, Drug_2) Drug_Pair_ttest = wilcox.test(Median.Mean..z.score ~ Drug_Number, data = Drug_Pair)\(p.value p_values = rbind(p_values, Drug_Pair_ttest) } } p_values <- as.vector(p_values) p_value_matrix <- matrix(p_values,nrow = length(unique(Neo.LY.Drug.Treatment.Data\)Drug)),ncol = length(unique(Neo.LY.Drug.Treatment.Data\(Drug))) columns <- as.vector(unique(Neo.LY.Drug.Treatment.Data\)Drug)) rows <- as.vector(unique(Neo.LY.Drug.Treatment.Data\(Drug)) p_value_dataframe <- as.data.frame(p_value_matrix, row.names = rows) names(p_value_dataframe) <- columns # print(p_value_dataframe) # Introduce Holm (also called Holm-Bonferroni) p-value correction (used because it gives same stringency against false positives (type 1 errors) with lower probability of false negatives (type 2 errors))) corrected_p_values <- p.adjust(p_values, method = \holm\, n = length(p_values)) corrected_p_value_matrix3 <- matrix(corrected_p_values,nrow = length(unique(Neo.LY.Drug.Treatment.Data\)Drug)),ncol = length(unique(Neo.LY.Drug.Treatment.Data\(Drug))) columns <- as.vector(unique(Neo.LY.Drug.Treatment.Data\)Drug)) rows <- as.vector(unique(Neo.LY.Drug.Treatment.Data$Drug)) corrected_p_value_dataframe <- as.data.frame(corrected_p_value_matrix3, row.names = rows) names(corrected_p_value_dataframe) <- columns print(corrected_p_value_dataframe)
Mann-Whitney test for significant Drug treatment effects on Focal adhesion intensity
rr p_values <- c() for (i in 1:length(unique(Neo.LY.Drug.Treatment.Data\(Drug))){ for (j in 1:length(unique(Neo.LY.Drug.Treatment.Data\)Drug))){ Drug_1 = subset(Neo.LY.Drug.Treatment.Data, Neo.LY.Drug.Treatment.Data\(Drug == unique(Neo.LY.Drug.Treatment.Data\)Drug)[i]) Drug_1\(Drug_Number = as.factor(i) Drug_2 = subset(Neo.LY.Drug.Treatment.Data, Neo.LY.Drug.Treatment.Data\)Drug == unique(Neo.LY.Drug.Treatment.Data\(Drug)[j]) Drug_2\)Drug_Number = as.factor(j+100) Drug_Pair = rbind(Drug_1, Drug_2) Drug_Pair_ttest = wilcox.test(Median.Mean._T.CMACs.z.score ~ Drug_Number, data = Drug_Pair)\(p.value p_values = rbind(p_values, Drug_Pair_ttest) } } p_values <- as.vector(p_values) p_value_matrix <- matrix(p_values,nrow = length(unique(Neo.LY.Drug.Treatment.Data\)Drug)),ncol = length(unique(Neo.LY.Drug.Treatment.Data\(Drug))) columns <- as.vector(unique(Neo.LY.Drug.Treatment.Data\)Drug)) rows <- as.vector(unique(Neo.LY.Drug.Treatment.Data\(Drug)) p_value_dataframe <- as.data.frame(p_value_matrix, row.names = rows) names(p_value_dataframe) <- columns # print(p_value_dataframe) # Introduce Holm (also called Holm-Bonferroni) p-value correction (used because it gives same stringency against false positives (type 1 errors) with lower probability of false negatives (type 2 errors))) corrected_p_values <- p.adjust(p_values, method = \holm\, n = length(p_values)) corrected_p_value_matrix4 <- matrix(corrected_p_values,nrow = length(unique(Neo.LY.Drug.Treatment.Data\)Drug)),ncol = length(unique(Neo.LY.Drug.Treatment.Data\(Drug))) columns <- as.vector(unique(Neo.LY.Drug.Treatment.Data\)Drug)) rows <- as.vector(unique(Neo.LY.Drug.Treatment.Data$Drug)) corrected_p_value_dataframe <- as.data.frame(corrected_p_value_matrix4, row.names = rows) names(corrected_p_value_dataframe) <- columns print(corrected_p_value_dataframe)
Mann-Whitney test for significant Drug treatment effects on intensity ratio Z-score
rr p_values <- c() for (i in 1:length(unique(Neo.LY.Drug.Treatment.Data\(Drug))){ for (j in 1:length(unique(Neo.LY.Drug.Treatment.Data\)Drug))){ Drug_1 = subset(Neo.LY.Drug.Treatment.Data, Neo.LY.Drug.Treatment.Data\(Drug == unique(Neo.LY.Drug.Treatment.Data\)Drug)[i]) Drug_1\(Drug_Number = as.factor(i) Drug_2 = subset(Neo.LY.Drug.Treatment.Data, Neo.LY.Drug.Treatment.Data\)Drug == unique(Neo.LY.Drug.Treatment.Data\(Drug)[j]) Drug_2\)Drug_Number = as.factor(j+100) Drug_Pair = rbind(Drug_1, Drug_2) Drug_Pair_ttest = wilcox.test(as.numeric(Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score) ~ Drug_Number, data = Drug_Pair)\(p.value p_values = rbind(p_values, Drug_Pair_ttest) } } p_values <- as.vector(p_values) p_value_matrix <- matrix(p_values,nrow = length(unique(Neo.LY.Drug.Treatment.Data\)Drug)),ncol = length(unique(Neo.LY.Drug.Treatment.Data\(Drug))) columns <- as.vector(unique(Neo.LY.Drug.Treatment.Data\)Drug)) rows <- as.vector(unique(Neo.LY.Drug.Treatment.Data\(Drug)) p_value_dataframe <- as.data.frame(p_value_matrix, row.names = rows) names(p_value_dataframe) <- columns # print(p_value_dataframe) # Introduce Holm (also called Holm-Bonferroni) p-value correction (used because it gives same stringency against false positives (type 1 errors) with lower probability of false negatives (type 2 errors))) corrected_p_values <- p.adjust(p_values, method = \holm\, n = length(p_values)) corrected_p_value_matrix5 <- matrix(corrected_p_values,nrow = length(unique(Neo.LY.Drug.Treatment.Data\)Drug)),ncol = length(unique(Neo.LY.Drug.Treatment.Data\(Drug))) columns <- as.vector(unique(Neo.LY.Drug.Treatment.Data\)Drug)) rows <- as.vector(unique(Neo.LY.Drug.Treatment.Data$Drug)) corrected_p_value_dataframe <- as.data.frame(corrected_p_value_matrix5, row.names = rows) names(corrected_p_value_dataframe) <- columns print(corrected_p_value_dataframe)
Figure 6
Import Data for Figure 6
rr # Set your path to input data data_input_path <- /Users/johnlock/Dropbox/A-CMAC Manchester collaboration/NCB submission documents/Resubmission 2/Source_Data
data_file <- Table 1 Statistics Source Data.xlsx
# Import individual data files Proliferation <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 6A, col_names = TRUE) Proliferation\(Day <- as.factor(Proliferation\)Day) EDU_Incorporation_Percentage <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 6B, col_names = TRUE) EDU_Incorporation_Percentage <- EDU_Incorporation_Percentage[,1:2]
Summarise Proliferation data per Condition per Day
rr Prol_Sum <- Proliferation %>% group_by(.dots = c(‘Condition’, ‘Day’)) %>% summarise( mean_proliferation = mean(Relative.Proliferation, na.rm = TRUE), SD_proliferation = sd(Relative.Proliferation, na.rm = TRUE) )
Plot Cell Proliferation Comparison of control vs b5KD
rr proliferation_linechart_filename <- 6A proliferation linechart
# The errorbars overlapped, so use position_dodge to move them horizontally pd <- position_dodge(0.1) # move them .05 to the left and right svglite(paste(panel_output_path, proliferation_linechart_filename, .svg, sep = \), width=4, height=3) ggplot(Prol_Sum, aes(x=Day, y=mean_proliferation, colour=Condition, group=Condition)) + geom_errorbar(aes(ymin=mean_proliferation-SD_proliferation, ymax=mean_proliferation+SD_proliferation), width=.1, position=pd) + scale_color_manual(values = c( = #49494A, 5KD = #EC1C24)) + geom_line(position=pd, size = 1) + geom_point(position=pd, size=3, shape=19) + # 21 is filled circle xlab() + ylab(Proliferation) + theme_light() dev.off()
null device
1 rr ggplot(Prol_Sum, aes(x=Day, y=mean_proliferation, colour=Condition, group=Condition)) + geom_errorbar(aes(ymin=mean_proliferation-SD_proliferation, ymax=mean_proliferation+SD_proliferation), width=.1, position=pd) + scale_color_manual(values = c( = #49494A, 5KD = #EC1C24)) + geom_line(position=pd, size = 1) + geom_point(position=pd, size=3, shape=19) + # 21 is filled circle xlab() + ylab(Proliferation) + theme_light()
T-testing Cell Proliferation Comparison per Condition per Day
rr Proliferation\(Day <- as.factor(Proliferation\)Day) p_values <- c() for (i in 1:length(unique(Proliferation$Day))){
single_day_data = subset(Proliferation, Proliferation$Day == unique(Proliferation$Day)[i])
Control_vs_b5KD_per_day_ttest = t.test(Relative.Proliferation ~ Condition, data = single_day_data, paired = FALSE)$p.value
p_values = rbind(p_values, Control_vs_b5KD_per_day_ttest)} p_values <- as.vector(p_values) p_value_matrix <- matrix(p_values,nrow = 1,ncol = length(p_values)) columns <- as.vector(paste(, unique(Proliferation\(Day), sep = \ \)) rows <- paste(unique(Proliferation\)Condition)[1], unique(Proliferation$Condition)[2], sep = _vs_) p_value_dataframe <- as.data.frame(p_value_matrix, row.names = rows) names(p_value_dataframe) <- columns print(p_value_dataframe)
EDU Incorporation Comparison per Condition
Fig 6B Plot EDU Incorporation Boxplot per Condition
rr #define panel_file_name EDU_boxplot_panel_file_name <- 6B EDU Incorporation Boxplots
EDU_Incorporation_Percentage\(Condition <- factor(EDU_Incorporation_Percentage\)Condition, c(, 5KD)) svglite(paste(panel_output_path, EDU_boxplot_panel_file_name, .svg, sep = \), width=4, height=6) boxplot(Percentage.EDU.positive.Cells.per.image ~ Condition, data = EDU_Incorporation_Percentage, notch = T, varwidth = T, outline = F, ylab = % EDU positive cells, col = c(control = #49494A, b5KD = #EC1C24), ylim = c(0,115)) beeswarm(Percentage.EDU.positive.Cells.per.image ~ Condition, data = EDU_Incorporation_Percentage, method = ‘center’, add = T, col = 2, pch = 1, cex = 1.3) dev.off()
null device
1 rr boxplot(Percentage.EDU.positive.Cells.per.image ~ Condition, data = EDU_Incorporation_Percentage, notch = T, varwidth = T, outline = F, ylab = % EDU positive cells, col = c(control = #49494A, b5KD = #EC1C24), ylim = c(0,115)) r beeswarm(Percentage.EDU.positive.Cells.per.image ~ Condition, data = EDU_Incorporation_Percentage, method = ‘center’, add = T, col = 2, pch = 1, cex = 1.3)
Fig6B Mann-Whitney µ test for EDU Incorporation differences
rr #Note: Equivalent to Wilcoxon rank sum test as data is unpaired Utest.b5KD.vs.control_EDU <- wilcox.test(Percentage.EDU.positive.Cells.per.image ~ Condition, data = EDU_Incorporation_Percentage)
cannot compute exact p-value with tiesrr Mann_Whitney_EDU_Incorporation_Score <- Utest.b5KD.vs.control_EDU$p.value print(Mann_Whitney_EDU_Incorporation_Score)
[1] 0.5337576Figure 7
Import Data for Figure 7
rr # Set your path to input data data_input_path <- /Users/johnlock/Dropbox/A-CMAC Manchester collaboration/NCB submission documents/Resubmission 2/Source_Data
data_file <- Table 1 Statistics Source Data.xlsx
# Import individual data files STORM.Data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 7E_F, col_names = TRUE)
Analysis of mitotic STORM data
STORM data wrangling2
rr ## Determine CMAC.type (Reticular vs Focal depending on presence of A in \(File (column)) CMAC.type = c() for (cluster in 1:length(STORM.Data\)File) ) { CMAC.type[cluster] <- if(grepl(, STORM.Data\(File[cluster]) == TRUE) { \Reticular\
} else if(grepl(\T\, STORM.Data\)File[cluster]) == TRUE) {
} else if(grepl(, STORM.Data\(File[cluster]) == TRUE) { \Non-Retraction\
} else { \Retraction\
} } STORM.Data\)CMAC.type <- cbind(as.character(CMAC.type)) ## Unique nanocluster ID generation = merge of Folder, File and Cluster STORM.Data\(Unique.ncID <- as.factor(paste(STORM.Data\)Folder, STORM.Data\(File, STORM.Data\)Cluster)) STORM.Data\(Unique.CMACID <- as.factor(paste(STORM.Data\)Folder, STORM.Data$File))
Nearest Neighbour Analysis between localisations within unique nanoclusters2
rr ## NND analysis of Incite3 data NNND <- c() for (i in 1:length(unique(STORM.Data\(Unique.CMACID)) ) { temp = STORM.Data[STORM.Data\)Unique.CMACID == unique(STORM.Data\(Unique.CMACID)[i], ] NNND = append(NNND, nndist(temp\)Centroid.nm., temp$Centroid.nm..1, k=1)) }
NAs introduced by coercionNAs introduced by coercionrr STORM.Data$NNND <- cbind(NNND)
Boxplots summarising Nearest Neighbout Distance per nanocluster2
rr NNND.A = subset(STORM.Data\(NNND, CMAC.type == \Non-Retraction\) NNND.T = subset(STORM.Data\)NNND, CMAC.type == ) TEST_nnnd_NRvsR <- wilcox.test(NNND.A, NNND.T, paired = FALSE) NNND_boxplot_filename <- 7E NNND boxplot
svglite(paste(panel_output_path, NNND_boxplot_filename, .svg, sep = \), width=4, height=3) boxplot(NNND ~ CMAC.type, subset(STORM.Data, CMAC.type == -Retraction | CMAC.type == ), notch = TRUE, outline = FALSE, main = Neighbour Distance of Nanocluster by CMAC Type, sub = TEST_nnnd_NRvsR$p.value, xlab = type, ylab = Neighbour Distance (nm), plot =
) dev.off()
null device
1 rr boxplot(NNND ~ CMAC.type, subset(STORM.Data, CMAC.type == -Retraction | CMAC.type == ), notch = TRUE, outline = FALSE, main = Neighbour Distance of Nanocluster by CMAC Type, sub = TEST_nnnd_NRvsR$p.value, xlab = type, ylab = Neighbour Distance (nm), plot =
)
Boxplots summarising Molecular Localization per nanocluster2
rr STORM.Data\(Molecules <- as.numeric(STORM.Data\)Molecules)
NAs introduced by coercionrr Molecules.A = subset(STORM.Data\(Molecules, CMAC.type == \Non-Retraction\) Molecules.T = subset(STORM.Data\)Molecules, CMAC.type == ) Mol_Count_boxplot_filename <- 7F Molecule Count boxplot
svglite(paste(panel_output_path, Mol_Count_boxplot_filename, .svg, sep = \), width=4, height=3) TEST_Molecules.A_RetvsFoc <- wilcox.test(Molecules.A, Molecules.T, paired = FALSE) boxplot(Molecules ~ CMAC.type, subset(STORM.Data, CMAC.type == -Retraction | CMAC.type == ), notch = TRUE,outline = FALSE, main = of Integrin avb5 Molecules by CMAC Type, sub = TEST_Molecules.A_RetvsFoc$p.value, xlab = type, ylab = of Integrin avb5 Molecules, plot =
) dev.off()
null device
1 rr TEST_Molecules.A_RetvsFoc <- wilcox.test(Molecules.A, Molecules.T, paired = FALSE) boxplot(Molecules ~ CMAC.type, subset(STORM.Data, CMAC.type == -Retraction | CMAC.type == ), notch = TRUE,outline = FALSE, main = of Integrin avb5 Molecules by CMAC Type, sub = TEST_Molecules.A_RetvsFoc$p.value, xlab = type, ylab = of Integrin avb5 Molecules, plot =
)
Figure 8
Import Data for Figure 8
rr # Set your path to input data data_input_path <- /Users/johnlock/Dropbox/A-CMAC Manchester collaboration/NCB submission documents/Resubmission 2/Source_Data
data_file <- Table 1 Statistics Source Data.xlsx
# Import individual data files Residual_Angle_data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 8A_B, col_names = TRUE) Cell_Division_Defect_Quantification <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 8C, col_names = TRUE) Cell_Division_Defect_Quantification <- Cell_Division_Defect_Quantification[1:9, 1:4]
Residual Angle Comparisons per Condition
Fig 8A Plot Residual Angle Boxplot of +/- Beeswarm by condition
rr #define panel_file_name boxplot_panel_file_name <- 8A Residual Angle Boxplots
# oldpar <- par() # par(mfrow = c(1,1), mar = c(5,5,1,1)) Residual_Angle_dataCN <- Residual_Angle_data # Set Condition order Residual_Angle_dataCN\(Condition <- factor(Residual_Angle_dataCN\)Condition, c(, 5KD, )) svglite(paste(panel_output_path, boxplot_panel_file_name, .svg, sep = \), width=4, height=6) boxplot(Pre.to.Post.Mitosis.Angle ~ Condition, data = Residual_Angle_dataCN, notch = T, varwidth = T, outline = F, ylab = Angle, col = c(control = #49494A, b5KD = #EC1C24, Rescue = #BCBCBC), ylim = c(0,115)) beeswarm(Pre.to.Post.Mitosis.Angle ~ Condition, data = Residual_Angle_dataCN, method = ‘center’, add = T, col = 2, pch = 1, cex = 0.5) dev.off()
null device
1 rr boxplot(Pre.to.Post.Mitosis.Angle ~ Condition, data = Residual_Angle_dataCN, notch = T, varwidth = T, outline = F, ylab = Angle, col = c(control = #49494A, b5KD = #EC1C24, Rescue = #BCBCBC), ylim = c(0,115)) r beeswarm(Pre.to.Post.Mitosis.Angle ~ Condition, data = Residual_Angle_dataCN, method = ‘center’, add = T, col = 2, pch = 1, cex = 0.5)
Fig 8B Plot Residual Angle smoothed density plot per condition
rr #define panel_file_name density_panel_file_name <- 8B Residual Angle Density Plot
svglite(paste(panel_output_path, density_panel_file_name, .svg, sep = \), width=4, height=6) sm.density.compare(Residual_Angle_dataCN\(Pre.to.Post.Mitosis.Angle, Residual_Angle_dataCN\)Condition, col = c(#49494A, #EC1C24, #BCBCBC), lty = c(1,1,1), lwd = 2, h = 5, xlim = c(0,90), xlab = Angle) colfill<-c(#49494A, #EC1C24, #BCBCBC) legend(x=40, y=0.033, levels(Residual_Angle_dataCN$Condition), fill=colfill, bty = ) dev.off()
null device
1 rr sm.density.compare(Residual_Angle_dataCN\(Pre.to.Post.Mitosis.Angle, Residual_Angle_dataCN\)Condition, col = c(#49494A, #EC1C24, #BCBCBC), lty = c(1,1,1), lwd = 2, h = 5, xlim = c(0,90), xlab = Angle) r colfill<-c(#49494A, #EC1C24, #BCBCBC) legend(x=40, y=0.033, levels(Residual_Angle_dataCN$Condition), fill=colfill, bty = )
Count cell number per condition
rr Residual_Angle_dataCN %>% group_by(Condition) %>% summarise( n = n())
Mann-Whitney µ test for Residual Angle differences
rr #Note: Equivalent to Wilcoxon rank sum test as data is unpaired b5KD_Control <- subset(Residual_Angle_data, Residual_Angle_data\(Condition == \b5KD\ | Residual_Angle_data\)Condition == ) Utest.b5KD.vs.control <- wilcox.test(Pre.to.Post.Mitosis.Angle ~ Condition, data = b5KD_Control) Utest.b5KD.vs.control.pval <- Utest.b5KD.vs.control\(p.value b5KD_Rescue <- subset(Residual_Angle_data, Residual_Angle_data\)Condition == 5KD | Residual_Angle_data\(Condition == \Rescue\) Utest.b5KD.vs.Rescue <- wilcox.test(Pre.to.Post.Mitosis.Angle ~ Condition, data = b5KD_Rescue) Utest.b5KD.vs.Rescue.pval <- Utest.b5KD.vs.Rescue\)p.value Control_Rescue <- subset(Residual_Angle_data, Residual_Angle_data\(Condition == \Control\ | Residual_Angle_data\)Condition == ) Utest.control.vs.Rescue <- wilcox.test(Pre.to.Post.Mitosis.Angle ~ Condition, data = Control_Rescue) Utest.control.vs.Rescue.pval <- Utest.control.vs.Rescue\(p.value Mann_Whitney_Scores <- data.frame(Condition_Comparison = c(\b5KD_vs_Control\, \b5KD_vs_Rescue\, \Control_vs_Rescue\)) Mann_Whitney_Scores\)pvalue <- rbind(Utest.b5KD.vs.control.pval, Utest.b5KD.vs.Rescue.pval, Utest.control.vs.Rescue.pval) print(Mann_Whitney_Scores)
Residual Angle Comparisons per Condition per Experiment
Plot Residual Angle Boxplot by condition by experiment
rr Residual_Angle_dataFN <- Residual_Angle_data Residual_Angle_dataFN\(File.Name <- factor(Residual_Angle_dataFN\)File.Name, c(_3_MMStack_control Hela 01.ome.tif, _3_MMStack_control Hela 02.ome.tif, _3_MMStack_b5 KD plus GFP 01.ome.tif, _3_MMStack_b5 KD plus GFP 02.ome.tif, _3_MMStack_b5 KD plus WTb5-GFP 01.ome.tif, _3_MMStack_b5 KD plus WTb5-GFP 02.ome.tif)) boxplot(Pre.to.Post.Mitosis.Angle ~ File.Name, data = Residual_Angle_dataFN, notch = T, varwidth = T, outline = F, ylab = Angle, col = c(#49494A, #49494A, #EC1C24, #EC1C24,#BCBCBC, #BCBCBC), ylim = c(0,115), names = c(_01, _02, 5KD_01, 5KD_02, _01, _02))
rr # beeswarm(Pre.to.Post.Mitosis.Angle ~ File.Name, data = Residual_Angle_dataFN, method = ‘swarm’, add = T, col = 2, pch = 0, cex = 0.7)
Plot Plot Residual Angle smoothed density plot per condition per experiment
rr sm.density.compare(Residual_Angle_dataFN\(Pre.to.Post.Mitosis.Angle, Residual_Angle_dataFN\)File.Name, col = c(#49494A, #49494A, #EC1C24, #EC1C24,#BCBCBC,#BCBCBC), h=5, lty = c(1,5,1,5,1,5), lwd = 2, xlim = c(0,90), xlab = Angle) legend(x=23, y= 0.05, levels(Residual_Angle_dataFN$File.Name), fill=c(#49494A, #49494A, #EC1C24, #EC1C24,#BCBCBC,#BCBCBC), bty = ‘n’)
Count cell number per condition per experiment
rr Residual_Angle_dataFN %>% group_by(File.Name) %>% summarise( n = n())
Cell Division Defect Quantification
Fig 8C Cell Division Defect Rate Plotting
rr CDD_boxplot_panel_file_name <- 8C Cell Division Defect Rate Boxplot
Cell_Division_Defect_Quantification\(Condition <- factor(Cell_Division_Defect_Quantification\)Condition, c(, 5KD, )) svglite(paste(panel_output_path, CDD_boxplot_panel_file_name, .svg, sep = \), width=4, height=6) boxplot(normal.cell.division ~ Condition, data = Cell_Division_Defect_Quantification, outline = FALSE, boxlty = 0, whisklty = 0, staplelty = 0, ylab = Division Percentage, ylim = c(0,115), medlwd = 0.0) beeswarm(normal.cell.division ~ Condition, data = Cell_Division_Defect_Quantification, method = ‘center’, add = T, col = c(control = #49494A, b5KD = #EC1C24, Rescue = #BCBCBC), ylim = c(0,115), pch = 19, cex = 1.3) dev.off()
null device
1 rr boxplot(normal.cell.division ~ Condition, data = Cell_Division_Defect_Quantification, outline = FALSE, boxlty = 0, whisklty = 0, staplelty = 0, ylab = Division Percentage, ylim = c(0,115), medlwd = 0.0) r beeswarm(normal.cell.division ~ Condition, data = Cell_Division_Defect_Quantification, method = ‘center’, add = T, col = c(control = #49494A, b5KD = #EC1C24, Rescue = #BCBCBC), ylim = c(0,115), pch = 19, cex = 1.3)
Cell Division Defect Summaries
rr Division_Defect_Summary <- Cell_Division_Defect_Quantification %>% group_by(Condition) %>% summarise( Normal_Division = mean(normal.cell.division), Abnormal_Division = mean(abnormal.cell.division) )
print(Division_Defect_Summary)
T-testing for Cell Division Defect Rate differences
rr b5KD_Control_CDD <- subset(Cell_Division_Defect_Quantification, Condition == 5KD | Condition == ) T_test_b5KD_Control_CDD <- t.test(normal.cell.division ~ Condition, data = b5KD_Control_CDD)\(p.value b5KD_Rescue_CDD <- subset(Cell_Division_Defect_Quantification, Condition == \b5KD\ | Condition == \Rescue\) T_test_b5KD_Rescue_CDD <- t.test(normal.cell.division ~ Condition, data = b5KD_Rescue_CDD)\)p.value Control_Rescue_CDD <- subset(Cell_Division_Defect_Quantification, Condition == | Condition == ) T_test_Control_Rescue_CDD <- t.test(normal.cell.division ~ Condition, data = Control_Rescue_CDD)\(p.value Ttest_Scores_CDD <- data.frame(Condition_Comparison = c(\b5KD_Control_CDD\, \b5KD_Rescue_CDD\, \Control_Rescue_CDD\)) Ttest_Scores_CDD\)pvalue <- rbind(T_test_b5KD_Control_CDD, T_test_b5KD_Rescue_CDD, T_test_Control_Rescue_CDD) print(Ttest_Scores_CDD)
Supp Fig 5
Import Data Supp Fig 5
rr # Set your path to input data data_input_path <- /Users/johnlock/Dropbox/A-CMAC Manchester collaboration/NCB submission documents/Resubmission 2/Source_Data
data_file <- Table 1 Statistics Source Data.xlsx
# Import individual data files Arp23_Inhibition_Data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 5C, col_names = TRUE) Arp23_Inhibition_Data\(Drug <- recode(Arp23_Inhibition_Data\)Drug, Arp_Inhib_CTRL_689 = _CTRL_689)
Analyse Arp2/3 inhibitor responses relative to control
Boxplot changes in reticular adhesion b5 intensity2
rr Arp23_b5_intensity_boxplots_filename <- SF5C Arp23_b5_intensity_boxplots
cell.numbers.ARP23.screen <- count(Arp23_Inhibition_Data, Drug) count(Arp23_Inhibition_Data, Drug) r mean.cell.numbers.ARP23.screen <- summarise(cell.numbers.ARP23.screen, avg = mean(n), SD = sd(n), sum(n)) summarise(cell.numbers.ARP23.screen, avg = mean(n), SD = sd(n), sum(n))
rr svglite(paste(panel_output_path, Arp23_b5_intensity_boxplots_filename, .svg, sep = \), width=7, height=5) p1 <- ggplot(Arp23_Inhibition_Data, aes(x=Drug, y=Median.Mean..z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(_CTRL_689, _Inhib_666)) + theme(axis.text.x = element_text(angle=45, hjust = 1)) + ylim(-3, 6) + ylab(Reticular Adhesion Intensity) + theme(legend.position=) ## plot T-CMAC intensity Z-score ggplot2 boxplot - per Drug p2 <- ggplot(Arp23_Inhibition_Data, aes(x=Drug, y=Median.Mean._T.CMACs.z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(_CTRL_689, _Inhib_666)) + theme(axis.text.x = element_text(angle=45, hjust = 1)) + ylim(-3, 6) + ylab(Focal Adhesion Intensity) + theme(legend.position=) ## plot intensity ratio Z-score ggplot2 boxplot - per Drug p3 <- ggplot(Arp23_Inhibition_Data, aes(x=Drug, y=Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(_CTRL_689, _Inhib_666)) + theme(axis.text.x = element_text(angle=45, hjust = 1)) + ylim(-3, 6) + ylab(Reticular:Focal Adhesion Intensity) + theme(legend.position=) grid.arrange(p1, p2, p3, ncol = 3) dev.off()
null device
1 rr p1 <- ggplot(Arp23_Inhibition_Data, aes(x=Drug, y=Median.Mean..z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(_CTRL_689, _Inhib_666)) + theme(axis.text.x = element_text(angle=45, hjust = 1)) + ylim(-3, 6) + ylab(Reticular Adhesion Intensity) + theme(legend.position=) ## plot T-CMAC intensity Z-score ggplot2 boxplot - per Drug p2 <- ggplot(Arp23_Inhibition_Data, aes(x=Drug, y=Median.Mean._T.CMACs.z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(_CTRL_689, _Inhib_666)) + theme(axis.text.x = element_text(angle=45, hjust = 1)) + ylim(-3, 6) + ylab(Focal Adhesion Intensity) + theme(legend.position=) ## plot intensity ratio Z-score ggplot2 boxplot - per Drug p3 <- ggplot(Arp23_Inhibition_Data, aes(x=Drug, y=Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score, color = factor(Drug))) + geom_boxplot(notch = TRUE, outlier.colour = ) + scale_x_discrete(limits=c(_CTRL_689, _Inhib_666)) + theme(axis.text.x = element_text(angle=45, hjust = 1)) + ylim(-3, 6) + ylab(Reticular:Focal Adhesion Intensity) + theme(legend.position=) grid.arrange(p1, p2, p3, ncol = 3)
Mann-Whitney test for significant Drug treatment effects on Reticular adhesion intensity2
rr p_values <- c() for (i in 1:length(unique(Arp23_Inhibition_Data\(Drug))){ for (j in 1:length(unique(Arp23_Inhibition_Data\)Drug))){ Drug_1 = subset(Arp23_Inhibition_Data, Arp23_Inhibition_Data\(Drug == unique(Arp23_Inhibition_Data\)Drug)[i]) Drug_1\(Drug_Number = as.factor(i) Drug_2 = subset(Arp23_Inhibition_Data, Arp23_Inhibition_Data\)Drug == unique(Arp23_Inhibition_Data\(Drug)[j]) Drug_2\)Drug_Number = as.factor(j+100) Drug_Pair = rbind(Drug_1, Drug_2) Drug_Pair_ttest = wilcox.test(Median.Mean..z.score ~ Drug_Number, data = Drug_Pair)\(p.value p_values = rbind(p_values, Drug_Pair_ttest) } } p_values <- as.vector(p_values) p_value_matrix <- matrix(p_values,nrow = length(unique(Arp23_Inhibition_Data\)Drug)),ncol = length(unique(Arp23_Inhibition_Data\(Drug))) columns <- as.vector(unique(Arp23_Inhibition_Data\)Drug)) rows <- as.vector(unique(Arp23_Inhibition_Data\(Drug)) p_value_dataframe <- as.data.frame(p_value_matrix, row.names = rows) names(p_value_dataframe) <- columns # print(p_value_dataframe) # Introduce Holm (also called Holm-Bonferroni) p-value correction (used because it gives same stringency against false positives (type 1 errors) with lower probability of false negatives (type 2 errors))) corrected_p_values <- p.adjust(p_values, method = \holm\, n = length(p_values)) corrected_p_value_matrix6 <- matrix(corrected_p_values,nrow = length(unique(Arp23_Inhibition_Data\)Drug)),ncol = length(unique(Arp23_Inhibition_Data\(Drug))) columns <- as.vector(unique(Arp23_Inhibition_Data\)Drug)) rows <- as.vector(unique(Arp23_Inhibition_Data$Drug)) corrected_p_value_dataframe <- as.data.frame(corrected_p_value_matrix6, row.names = rows) names(corrected_p_value_dataframe) <- columns print(corrected_p_value_dataframe)
Mann-Whitney test for significant Drug treatment effects on Focal adhesion intensity2
rr p_values <- c() for (i in 1:length(unique(Arp23_Inhibition_Data\(Drug))){ for (j in 1:length(unique(Arp23_Inhibition_Data\)Drug))){ Drug_1 = subset(Arp23_Inhibition_Data, Arp23_Inhibition_Data\(Drug == unique(Arp23_Inhibition_Data\)Drug)[i]) Drug_1\(Drug_Number = as.factor(i) Drug_2 = subset(Arp23_Inhibition_Data, Arp23_Inhibition_Data\)Drug == unique(Arp23_Inhibition_Data\(Drug)[j]) Drug_2\)Drug_Number = as.factor(j+100) Drug_Pair = rbind(Drug_1, Drug_2) Drug_Pair_ttest = wilcox.test(Median.Mean._T.CMACs.z.score ~ Drug_Number, data = Drug_Pair)\(p.value p_values = rbind(p_values, Drug_Pair_ttest) } } p_values <- as.vector(p_values) p_value_matrix <- matrix(p_values,nrow = length(unique(Arp23_Inhibition_Data\)Drug)),ncol = length(unique(Arp23_Inhibition_Data\(Drug))) columns <- as.vector(unique(Arp23_Inhibition_Data\)Drug)) rows <- as.vector(unique(Arp23_Inhibition_Data\(Drug)) p_value_dataframe <- as.data.frame(p_value_matrix, row.names = rows) names(p_value_dataframe) <- columns # print(p_value_dataframe) # Introduce Holm (also called Holm-Bonferroni) p-value correction (used because it gives same stringency against false positives (type 1 errors) with lower probability of false negatives (type 2 errors))) corrected_p_values <- p.adjust(p_values, method = \holm\, n = length(p_values)) corrected_p_value_matrix7 <- matrix(corrected_p_values,nrow = length(unique(Arp23_Inhibition_Data\)Drug)),ncol = length(unique(Arp23_Inhibition_Data\(Drug))) columns <- as.vector(unique(Arp23_Inhibition_Data\)Drug)) rows <- as.vector(unique(Arp23_Inhibition_Data$Drug)) corrected_p_value_dataframe <- as.data.frame(corrected_p_value_matrix7, row.names = rows) names(corrected_p_value_dataframe) <- columns print(corrected_p_value_dataframe)
Mann-Whitney test for significant Drug treatment effects on intensity ratio Z-score2
rr p_values <- c() for (i in 1:length(unique(Arp23_Inhibition_Data\(Drug))){ for (j in 1:length(unique(Arp23_Inhibition_Data\)Drug))){ Drug_1 = subset(Arp23_Inhibition_Data, Arp23_Inhibition_Data\(Drug == unique(Arp23_Inhibition_Data\)Drug)[i]) Drug_1\(Drug_Number = as.factor(i) Drug_2 = subset(Arp23_Inhibition_Data, Arp23_Inhibition_Data\)Drug == unique(Arp23_Inhibition_Data\(Drug)[j]) Drug_2\)Drug_Number = as.factor(j+100) Drug_Pair = rbind(Drug_1, Drug_2) Drug_Pair_ttest = wilcox.test(Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score ~ Drug_Number, data = Drug_Pair)\(p.value p_values = rbind(p_values, Drug_Pair_ttest) } } p_values <- as.vector(p_values) p_value_matrix <- matrix(p_values,nrow = length(unique(Arp23_Inhibition_Data\)Drug)),ncol = length(unique(Arp23_Inhibition_Data\(Drug))) columns <- as.vector(unique(Arp23_Inhibition_Data\)Drug)) rows <- as.vector(unique(Arp23_Inhibition_Data\(Drug)) p_value_dataframe <- as.data.frame(p_value_matrix, row.names = rows) names(p_value_dataframe) <- columns # print(p_value_dataframe) # Introduce Holm (also called Holm-Bonferroni) p-value correction (used because it gives same stringency against false positives (type 1 errors) with lower probability of false negatives (type 2 errors))) corrected_p_values <- p.adjust(p_values, method = \holm\, n = length(p_values)) corrected_p_value_matrix8 <- matrix(corrected_p_values,nrow = length(unique(Arp23_Inhibition_Data\)Drug)),ncol = length(unique(Arp23_Inhibition_Data\(Drug))) columns <- as.vector(unique(Arp23_Inhibition_Data\)Drug)) rows <- as.vector(unique(Arp23_Inhibition_Data$Drug)) corrected_p_value_dataframe <- as.data.frame(corrected_p_value_matrix8, row.names = rows) names(corrected_p_value_dataframe) <- columns print(corrected_p_value_dataframe)
Supp Fig 6
Import Data Supp Fig 6
rr # Set your path to input data data_input_path <- /Users/johnlock/Dropbox/A-CMAC Manchester collaboration/NCB submission documents/Resubmission 2/Source_Data
data_file <- Table 1 Statistics Source Data.xlsx
# Import individual data files Secondary.siRNA.Screen.Data <- read_xlsx(paste(data_input_path, data_file, sep = /), sheet = 6A, col_names = TRUE)
Analyse Secondary siRNA Screen responses
Boxplot changes in reticular to focal adhesion b5 intensity ratio
rr Secondary_siRNA_Screen_boxplots_filename <- SF6A Secondary_siRNA_Screen_boxplot_summary
Secondary.siRNA.Screen.Data\(siRNA_Target <- recode(Secondary.siRNA.Screen.Data\)siRNA_Target, PIK3C2a_1 = 3C2A_1) Secondary.siRNA.Screen.Data\(siRNA_Target <- recode(Secondary.siRNA.Screen.Data\)siRNA_Target, PIK3C2a_2 = 3C2A_2) Secondary.siRNA.Screen.Data\(siRNA_Target <- recode(Secondary.siRNA.Screen.Data\)siRNA_Target, PIK3C2a_3 = 3C2A_3) Secondary.siRNA.Screen.Data\(siRNA_Target <- recode(Secondary.siRNA.Screen.Data\)siRNA_Target, PIK3C2a_4 = 3C2A_4) cell.numbers.secondary.screen <- count(Secondary.siRNA.Screen.Data, siRNA_Target) count(Secondary.siRNA.Screen.Data, siRNA_Target) r mean.cell.numbers.secondary.screen <- summarise(cell.numbers.secondary.screen, avg = mean(n), SD = sd(n), sum(n)) summarise(cell.numbers.secondary.screen, avg = mean(n), SD = sd(n), sum(n))
rr svglite(paste(panel_output_path, Secondary_siRNA_Screen_boxplots_filename, .svg, sep = \), width=7, height=5) ggplot(Secondary.siRNA.Screen.Data, aes(x=siRNA_Target, y=Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score, color = factor(siRNA_Treatment_Type))) + geom_boxplot(notch = TRUE, outlier.colour = ) + theme(axis.text.x = element_text(angle=90)) + ylim(-3, 6) + scale_x_discrete(limits=c(,,_3Ctrl_Pool, -TARGETplus Non-targeting Control, Non-targeting Control, , , 5, _1, _2, _3, _4, 4K2A_1, 4K2A_2, 4K2A_3, 4K2A_4, 4KA_1, 4KA_2, 4KA_3, 4KA_4, 5K1B_1, 5K1B_2, 5K1B_3, 5K1B_4, 5K1C_1, 5K1C_2, 5K1C_3, 5K1C_4, _1, _2, _3, _4, 3C2A_1, 3C2A_2, 3C2A_3, 3C2A_4)) dev.off()
null device
1 rr ggplot(Secondary.siRNA.Screen.Data, aes(x=siRNA_Target, y=Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score, color = factor(siRNA_Treatment_Type))) + geom_boxplot(notch = TRUE, outlier.colour = ) + theme(axis.text.x = element_text(angle=90)) + ylim(-3, 6) + scale_x_discrete(limits=c(,,_3Ctrl_Pool, -TARGETplus Non-targeting Control, Non-targeting Control, , , 5, _1, _2, _3, _4, 4K2A_1, 4K2A_2, 4K2A_3, 4K2A_4, 4KA_1, 4KA_2, 4KA_3, 4KA_4, 5K1B_1, 5K1B_2, 5K1B_3, 5K1B_4, 5K1C_1, 5K1C_2, 5K1C_3, 5K1C_4, _1, _2, _3, _4, 3C2A_1, 3C2A_2, 3C2A_3, 3C2A_4))
Mann-Whitney test for significant siRNA treatment effects2
rr siRNA.conditions.for.U.test <- subset(Secondary.siRNA.Screen.Data, siRNA_Target != & siRNA_Target != & siRNA_Target != _3Ctrl_Pool & siRNA_Target != Non-targeting Control) p_values <- c() for (i in 1:length(unique(siRNA.conditions.for.U.test\(siRNA_Target))){ for (j in 1:length(unique(siRNA.conditions.for.U.test\)siRNA_Target))){ siRNA_Target_1 = subset(siRNA.conditions.for.U.test, siRNA.conditions.for.U.test\(siRNA_Target == unique(siRNA.conditions.for.U.test\)siRNA_Target)[i]) siRNA_Target_1\(siRNA_Target_Number = as.factor(i) siRNA_Target_2 = subset(siRNA.conditions.for.U.test, siRNA.conditions.for.U.test\)siRNA_Target == unique(siRNA.conditions.for.U.test\(siRNA_Target)[j]) siRNA_Target_2\)siRNA_Target_Number = as.factor(j+100) siRNA_Target_Pair = rbind(siRNA_Target_1, siRNA_Target_2) siRNA_Target_Pair_ttest = wilcox.test(Ratio_A.CMAC.to.T.CMAC_mean_intensity.z.score ~ siRNA_Target_Number, data = siRNA_Target_Pair)\(p.value p_values = rbind(p_values, siRNA_Target_Pair_ttest) } } p_values <- as.vector(p_values) p_value_matrix <- matrix(p_values,nrow = length(unique(siRNA.conditions.for.U.test\)siRNA_Target)),ncol = length(unique(siRNA.conditions.for.U.test\(siRNA_Target))) columns <- as.vector(unique(siRNA.conditions.for.U.test\)siRNA_Target)) rows <- as.vector(unique(siRNA.conditions.for.U.test\(siRNA_Target)) p_value_dataframe <- as.data.frame(p_value_matrix, row.names = rows) names(p_value_dataframe) <- columns # print(p_value_dataframe) # Introduce Holm (also called Holm-Bonferroni) p-value correction (used because it gives same stringency against false positives (type 1 errors) with lower probability of false negatives (type 2 errors))) corrected_p_values <- p.adjust(p_values, method = \holm\, n = length(p_values)) corrected_p_value_matrix9 <- matrix(corrected_p_values,nrow = length(unique(siRNA.conditions.for.U.test\)siRNA_Target)),ncol = length(unique(siRNA.conditions.for.U.test\(siRNA_Target))) columns <- as.vector(unique(siRNA.conditions.for.U.test\)siRNA_Target)) rows <- as.vector(unique(siRNA.conditions.for.U.test$siRNA_Target)) corrected_p_value_dataframe <- as.data.frame(corrected_p_value_matrix9, row.names = rows) names(corrected_p_value_dataframe) <- columns print(corrected_p_value_dataframe)
Plot heatmap of corrected p_values2
rr heatmap.2(log(corrected_p_value_matrix9), dendrogram = ‘none’, Rowv = FALSE, Colv = FALSE, labRow = rows, labCol = columns, srtCol = 45, cexRow = 1, cexCol = 1, margins = c(7, 9), symm = TRUE, revC = FALSE, breaks = c(log(1e-300), log(1e-50), log(1e-10), log(1e-6), log(0.001), log(0.01), log(0.05), log(1)), rowsep = 1:nrow(corrected_p_value_matrix9), colsep = 1:ncol(corrected_p_value_matrix9), sepcolor = ‘darkgrey’, sepwidth = c(0.02,0.02), trace = ‘none’, col = c(, , , , , , ), denscol = NULL, keysize = 1.5, key.title = NA)
